Ecosystem engineers, such as beavers, facilitate other species through habitat modification. Beaver conservation and mimicry is rapidly growing as a restoration tool, but their ability to facilitate other species is poorly understood in many systems. Further, it is unknown if beaver dams uniquely benefit some functional groups (species with similar traits) more than others. We therefore examined the impact of beaver dams on fish communities in the northwestern Great Plains of the United States through two related studies, one observational and one experimental. In a survey of 30 streams, each with a paired beaver and control site, we found total and native fish abundance was higher in beaver complexes, but introduced species’ abundance was not. Functional group proportions within trophic, feeding, spawning, thermal, and habitat categories were not significantly different between beaver and control sites. We compared these results to those of an experimental, beaver-mimicry restoration in which 18 beaver-dam analogs were installed over three years on a prairie stream. Like the observational study, total fish abundance increased in the beaver-mimicry reach relative to a control reach. Functional group proportions changed slightly after initial restoration, but these differences were not sustained in the following year. Taken together, these studies suggest that beaver dams, by increasing wetted habitat, increase the total abundance of fish, but the proportional makeup of functional groups remains similar to the starting fish community. This serves as an example of how neutral theory can be applied to restoration and ecosystem engineering to better understand the mechanisms underlying assemblage-level responses.